An external penetration test shows what an attacker on the internet can reach. An internal test assumes they’re already inside and shows how far they could go. This guide explains what each covers, what each typically finds, which to do first and how to get ready.
TL;DR
- An external penetration test attacks you from the internet, the way an outsider would, and tests your firewall, VPN, remote access, email gateway, website and anything else exposed. An internal test assumes the attacker is already inside, such as on a phished laptop, and shows how far they could move and what they could reach.
- External tests usually find exposed services, unpatched edge devices and weak remote access. Internal tests usually find shared passwords, excessive admin rights, flat networks and paths to full domain control.
- If you’ve never tested, an external test is usually the first step. If phishing is your bigger worry, or a customer asks for evidence of internal controls, start with internal or do both together.
- They answer different questions, so most organisations need both over time.
In this post
Updated 3 November 2026 with figures from ASD’s Annual Cyber Threat Report 2025–26.
What is penetration testing?
Penetration testing (pen testing) is a controlled, authorised attack on your IT environment that checks for weaknesses an attacker could actually exploit. A tester works within an agreed scope and rules, proves which issues are real, and reports each one with evidence and a fix.
Tests can be grouped by where the attack starts. The two most common starting points are outside your network (external) and inside it (internal). Knowing the difference helps you choose the test that matches the risk you’re most worried about. For the full step-by-step process, from scoping to retest, see what happens in a penetration test.
What does an external penetration test cover?
An external test takes the position of a remote attacker with no special access. The goal is to find a way from the internet into your systems, or to data you didn’t mean to expose.
Typical scope:
- Firewalls and edge devices, including their management interfaces.
- VPNs and remote access portals, such as remote desktop gateways.
- Email gateways and other mail-facing services.
- Websites and web portals hosted on your addresses (a full web application test goes deeper; see below).
- Any other exposed service: file transfer, databases, camera systems, building management, printers, test servers.
The tester starts by mapping what’s visible: which addresses and domains belong to you, which services are listening and what versions they run. Then they test those services for known vulnerabilities, weak or default credentials, misconfiguration and ways around authentication.
What does an external test typically find?
- An edge device or VPN running firmware with a published vulnerability.
- Remote desktop or an admin interface exposed directly to the internet.
- A forgotten test or old server that nobody realised was still reachable.
- Login pages without MFA, or with no lockout, that allow password guessing.
- Outdated encryption settings on public services.
- Information leaking from error pages, file listings or DNS records that helps an attacker plan.
What does an internal penetration test cover?
An internal test simulates an attacker who is already inside your network. That might be malware on a staff laptop after a phishing email, a stolen device, a malicious insider or someone who plugs into a spare network port.
The testing techniques can be similar to an external test. The big difference is the assumption: the attacker has already got past your perimeter. To save time and mirror a real compromise, the organisation usually provides a standard user account or a laptop on the corporate network as the starting point.
Typical scope:
- Active Directory or Microsoft Entra ID: privileged groups, service accounts, password policy and the paths to domain administrator.
- Servers and file shares: who can read what, and where sensitive data is stored.
- Workstations: local admin rights, stored credentials, patch levels.
- Network segmentation: whether staff, servers, printers, cameras, guests and backups are actually kept apart.
- Backups: whether an attacker with ordinary access could reach and delete them.
What does an internal test typically find?
- The same local administrator password on every PC, so one compromised machine opens up the rest.
- Admin accounts used for everyday work, or signed in on ordinary laptops where their credentials can be stolen.
- Service accounts with weak passwords and far more privilege than they need.
- File shares open to "everyone" that hold HR, finance or client data.
- A flat network where any device can reach any other, including servers and backup storage.
- Legacy protocols left enabled that let an attacker capture or relay credentials.
These issues often look minor on their own. An internal test shows how they chain together. NIST’s SP 800-115 notes that most penetration tests look for "combinations of vulnerabilities" that give more access than any single one could.
This is also why an internal attack can do more damage than an external one. By the time an attacker is inside, perimeter defences such as the external firewall have already been bypassed.
Is "inside" really a likely starting point?
Yes. ASD’s Annual Cyber Threat Report 2025-26 maps the incidents it analysed to MITRE ATT&CK techniques. Phishing appeared in 39% of them and remained the most observed initial access technique. Valid accounts (an attacker signing in with real credentials) appeared in 15%, and user execution (someone opening a malicious file or link) in 17%. In other words, attackers very often get in through a person and a password, then work from inside. An internal test measures exactly that next step.
Internal vs external at a glance
| External penetration test | Internal penetration test | |
|---|---|---|
| Starting point | The internet, no access | Inside the network, usually as a standard user or device |
| Simulates | An outside attacker looking for a way in | A phished laptop, stolen device or insider |
| Main question | Can someone get in from outside? | Once in, how far can they get? |
| Typical targets | Firewall, VPN, remote access, email, exposed services | Active Directory, servers, shares, workstations, segmentation, backups |
| Typical findings | Unpatched edge devices, exposed admin interfaces, weak remote access | Shared passwords, excess privilege, flat network, reachable backups |
| What you need to provide | The list of addresses and domains in scope | A user account or device, network access, and the in-scope ranges |
| Fixes usually involve | Patching, closing services, adding MFA | Credential hygiene, admin separation, segmentation, hardening |
Which should you do first?
Pick the row that best describes you.
| Your situation | Start with | Why |
|---|---|---|
| You’ve never had a penetration test | External | It covers the doors every attacker on the internet can knock on, and the scope is usually small and clear |
| You have remote access, a VPN or self-hosted services on the internet | External | Edge devices are a common way in and are often patched late |
| Phishing is your main worry, or you’ve had a staff account compromised | Internal | It starts where a phished laptop would and shows the real blast radius |
| You run your own Active Directory and file servers | Internal | Paths to domain admin are what an attacker inside looks for first |
| Almost everything is in Microsoft 365 and cloud apps, with little on-premises | External plus a Microsoft 365 security review | There may be little internal network to test; sign-in and sharing settings matter more |
| A customer, insurer or government buyer asks for "a penetration test" | Ask which they mean, then match it | Questionnaires often specify external, internal or both |
| You’ve made major changes: new firewall, new office, merger, migration | Whichever side changed, or both | Testing confirms the change didn’t open a gap |
How do the two work together?
They’re two halves of one picture. The external test tells you how hard it is to get in. The internal test tells you how much damage follows if someone does. A strong perimeter with a weak inside means one successful phishing email can lead to everything. A well-hardened inside with a weak perimeter means you’re relying on detection to catch people who shouldn’t be there.
Many organisations run both in the same engagement, so one report shows the full path. A common pattern: the external test finds no way in, the internal test (beginning on a phished laptop) reaches the file server through a shared password, and the fixes for both are planned together.
Between tests, regular vulnerability scanning keeps known weaknesses from building up again.
Want to know what an attacker could actually do? Our penetration testing finds the weaknesses that matter in your network, cloud and applications, and shows you how to fix them.
How to prepare for each
For an external test:
- List every public IP address, domain and subdomain you believe is yours. Check: your IT provider and website host have confirmed the list.
- Get written agreement from anyone who hosts or manages in-scope systems. Check: cloud services are tested within the provider’s published rules (for example, Microsoft’s cloud penetration testing rules of engagement).
- Tell your security monitoring provider, unless detection is part of the test. Check: they know the dates and the tester’s source addresses.
For an internal test:
- Decide the starting point: a standard user account, a corporate laptop or a network port. Check: the account or device works before testing starts and has normal (not admin) rights.
- List the network ranges in scope and anything to avoid, such as medical equipment, building systems or a fragile legacy server. Check: exclusions are written into the rules of engagement.
- Confirm backups of in-scope servers. Check: a recent restore has been tested.
- Name a contact who can pause testing if something unexpected happens. Check: the tester has their mobile number.
For both: decide in advance who will own the fixes. Most internal findings, such as segmentation and admin separation, are design changes rather than patches, so they need planning. Our small business VLAN guide shows one way to segment a typical office.
Want your network set up properly? We design, build and secure business networks: segmentation, firewalls, Wi-Fi and remote access, documented so the next person understands it.
Key takeaways
- External testing asks whether an outsider can get in; internal testing asks how far an attacker gets once inside.
- External findings are usually exposed services and unpatched edge devices; internal findings are usually shared passwords, excess privilege and flat networks.
- Phishing and stolen credentials are among the most common ways in, which is why testing from a phished laptop matters.
- If you’ve never tested, start external; if phishing is your main worry or you run your own Active Directory, start internal or do both.
- Prepare with an accurate scope, written approvals, a working test account and a named contact.
Sources
- NIST, SP 800-115: Technical Guide to Information Security Testing and Assessment, section 5.2 Penetration Testing
- ASD’s ACSC, Annual Cyber Threat Report 2025-26, Figure 8: top 10 MITRE ATT&CK techniques
- Microsoft Security Response Center, Microsoft Cloud Penetration Testing Rules of Engagement
Want a second pair of eyes?
Catharsis Cyber Security is a Sydney cyber security company, certified to ISO/IEC 27001:2022. We help organisations protect, test and run their technology, and explain it in plain English.





Spawn a TTY Shell